<![CDATA[Diabetes mellitus (DM) is a chronic disease associated with high levels of sugar or glucose in the blood. Diabetes is caused by one of two causes, autoimmune reactions (the body's defense system attacks insulin-producing cells) or insulin resistance (the body does not fully respond to insulin). The purpose of this research is to create a machine learning model that can detect diabetes early. There are many ways to diagnose diabetes, one of the methods is using machine learning. Support Vector Machine (SVM) is a machine learning method that is known to be quite effective for classification cases. The dataset is cleaned and normalized before so it can be ready to input in the SVM model. The SVM model is processed and tested in order to find the best model for making a diagnosis. The output of the SVM model will diagnose patients who suffer diabetes or not. The SVM model is divided into two types, the benchmark model which is implemented using the Sequential Minimal Optimization (SMO) algorithm and the scratch model which is implemented using the Sequential Learning algorithm. Each model is optimized using the Grid Search algorithm so that it can find optimal hyperparameters that can be used by the model. The optimal model is retested on several metrics using 10-fold cross validation. The test results show that the benchmark model has 0,87 mean accuracy, 0,82 mean precision, 0,78 mean sensitivity, and 0,92 mean specificity. The scratch model has 0,78 mean accuracy, 0,69 mean precision, 0,59 mean sensitivity, dan 0,87 mean specificity. The experimental results show that the Support Vector Machine method has the potential to be used as an early detection tool for diabetes.]]>
Copyrights © 2021
